How it works
There is no proprietary index here and no cached availability list. Each result is a live question put to whoever actually knows — the registry, the platform, the open web — and reported back with the source attached.
Whatever you type — a brand name with spaces, a pasted URL, a domain — is reduced to the label a registrar and a package registry would both accept. We show you exactly what we checked, so there is never a gap between what you asked and what we answered.
IANA publishes a map of which RDAP endpoint serves which TLD. We resolve your TLD through that map and query the registry directly. A record means registered; no record means available. Anything else — a timeout, an odd status, a body that isn't an RDAP record — is reported as unknown rather than rounded to the answer you'd prefer.
Each handle is your name run through that platform's own rules, then fetched. GitHub, X, YouTube, npm, PyPI and crates.io answer differently for a taken handle and a free one, which is the only property that makes a probe worth running. You get the exact handle we tried and a link to see it yourself.
Two brand-shaped searches pull public pages — brand sites, company registers, product listings. A model extracts the brands it can see in them, and every citation is checked against the pages we actually fetched: an invented register link is deleted before you see it. The risk band is then a fixed rule over what survived.
The surface
Ask for any TLD you like, up to twelve a check. One whose registry publishes no RDAP service comes back unknown, with the reason written out.
.co — The .co registry publishes no RDAP service to IANA, and the endpoints that answer are not authoritative — one 404s for registered names, the other returns a help page for every label. We will not guess.
Each handle follows that platform’s own naming rules. Where a rule makes your name impossible — X caps at fifteen characters — we report the rule instead of a status we could not observe.
What we refuse
Six popular platforms are missing from this product on purpose. Each was asked about a handle that was free and one that was taken, and gave the same reply both times.
Asking a platform about a free handle and a taken one.
Returns 200 for handles that are free, so a hit means nothing.
Returns 200 for handles that are free, so a hit means nothing.
Returns 200 for handles that are free, so a hit means nothing.
Returns 403 to server-side callers regardless of the handle.
Returns 403 to server-side callers regardless of the handle.
Blocks server-side requests, so availability cannot be observed.
Trademark
The model never assigns the risk band. It reads the pages we fetched and names the brands it can see; anything it cites that we did not fetch is deleted. The band is then a fixed function of what survived, which you can re-derive yourself from the response.
Reading public web evidence for existing brands.
A low band is not a clearance.
We looked for public evidence of brands using your name and did not find much. That is a useful signal and it is not legal advice. The USPTO and EUIPO registers were never queried — there is no keyless way to search them — so before you commit, have a trademark attorney search the registers in the markets you sell into.